Round tubes are back.

The goal of this build has various missions:

1) show how easy it is to build a ship on par with commercial ships.

2) stay aloft for over 30 minutes carrying a full-size camera.

3) use low Kv motors and low voltage to create an efficient ship but stable camera platform (a design that is usually contrary to the objective).

4) fly at slightly less than 200 Hz but also be able to fly at 200 Hz (the evil frequency) to test the impact of doing that when vibrations are only managed by frame stiffness and low mass.

5) test light methods to mitigate the more fragile cloth/woven carbon tubes.

6) test APM code flexibility on flying a new octa motor layout.

After a successful test using braided carbon fiber tubes that are basically indestructible, strong, stiff, light, and provide a surface that is easy to bond, this discussion will focus on using cloth or woven carbon fiber tubes.  Relative to braided tubes, woven tubes have a smooth surface (not as easy to bond), have a lower resin content, and layered fibers. Braided tubes use fibers that are braided from the inside layer to the outside layer.  Woven tubes stack and bond either cloth or tape layers placed in multiple directions to derive engineered stiffness and light weight.  But unlike pultruded (where most all fibers in one direction), there are axial fibers to keep cracks from propagating. Why not just stay with braided? Woven carbon tubes are more available, come in more sizes, and are lighter for their stiffness.

This discussion will go from design through flight tests.

Views: 14078

Reply to This

Replies to This Discussion

Installment 1:  Requirements.  

Set some measurable criteria for your ship.  For this ship, mine were:

o An 80 degree clean (no props) field of view for a camera sensor located at the ship geometric center (full size camera).

o A 120 FOV with the camera sensor located less than 7" forward of center (go pro) for pan 90 degree downward to forward to 90 degrees upward.

o Total frame weight (bolts, masts, spars, platform, gussets, ties, etc.) < 400 grams to derive

   - stiffness resulting in x/y/z accels < .1g

   - not fail at 9 gravities x 3.2 kg (approx max ship load)

o Land safely with any 1-motor failure (octa?)

o Rotors with > 2 x 2.3 kg net-thrust (thrust after lifting themselves) or 575 net grams thrust per rotor 

o APM located at center of x/y CG and positioned at the propeller z-plane.

o Able to backpack.

o Able to assemble in field in < 12 minutes.

o Flight time with GoPro of > 30 minutes.

o Ability to support gimbal for high quality stills and videos.

o 140 Hz < hover frequency < 180 Hz.

Installment II:  Preliminary Design

Preliminary design can be as simple as looking at your desk top and just saying, "I'm going to try to meet my requirements using stuff in this pile."  Or it can be as time consuming as spending a few years or so testing motors, props, ESCs, batteries, structures, controllers, wiring, build methods, landing gear, camera gimbals, vibration systems, radios, etc.

So confession time.  I did the latter and still have a long way to go ... learning every day thanks to patient bloggers and friends.

To meet requirements, weight and lift efficiency were the two biggest factors.  A year of research bore out the following gains:

1) Rotor (motor, prop, mounts) net-lift to mass improvement of 30%

2) Structure (frame, platform, fasteners) 120 gram improvement

3) Electronics (components and wiring) 35 gram improvement

4) Batteries net duration to mass improvement of 3%

Rotors:  When through a list of over 2,000 motors comparing previous test results.  Then bought what appeared to be the best for test.


For the test also bought an assortment of propellers for each of the motors ( ... so a lot of combinations).

Made a multicopter test frame to measure watts and lift.  Then tested based on net-lift per watt.  Testing for multicopters is different than testing for planes.  A rotor has to lift itself first before it can benefit the rest of the ship, hence net-lift.  Obviously too, a larger prop takes more structure, so the rotor has to also lift additional frame mass (for my builds about 2 grams per inch per rotor for props over 10").  In any case, to make a long story short, I ended up chosing an RCTimer HP4215-460KV motor with an unidentified 15 x 5.5 Carbon prop I found in China.  I went for the 460KV range to get the natural frequency of the ship just below 200 Hz during hover to meet that particular requirement.  Is this combination the optimal rotor?  Heck no.  Every time I think I have the best showing huge improvements, someone comes along and recommends something else.  Right now Hugues has me looking at T-Motors.

Structure:  Bought about every wood, aluminum, and carbon piece of structure out there (even archery arrows).  Tested for durability and stiffness to weight.  From flying kites knew that pultruded carbon sucked, so ended up focusing on wrapped, braided, and cloth carbon tubes.  Ended up with three potential tubes to be used in the design process, each ideal for different loads and lengths.  From left to right:

o  Braided carbon tube -      11mm OD   100% strength factor     0.87 g/cm  (indestructible)

o  Fabric carbon tube -        15mm OD   102% strength factor     0.69 g/cm  (moderate lengths and weights)

o  Wrapped carbon tube -    22mm OD    253% strength factor    0.84 g/cm  (long lengths and higher loads)

Platform:  The lightest/strength material out there for platforms is what you walk on when you fly - carbon sandwich panels.  A few of my patents were based around the use and optimization of these materials on Boeing and Airbus aircraft.  The downside is that these materials are expensive.  But they are avialable.  This panel is a 1/4" thick single ply carbon sheet on top and bottom that sandwiches a Nomex core (hex expanded epoxy impregnated paper).

Controller:  APM of course.  The open source makes custom copters possible.  And a fantastic set of intelligent and knowledgeable developers that are willing to share.

ESC:  Haven't done a lot of work in this area so just used one that works with APM.  Was drawn to the MDrive because you could program the precise advance timing and optimize efficiency.  But APM didn't work with them.

Wiring: The lightest wire I found was UL AWM Style 1429.  The sheath is UV protected and thin (light).  But it's not easy to find.  So pay attention to the amount of plastic on the wire.  You don't need much.  The copper itself obviously needs to be appropriately sized in the design process to make the ESC to motor runs.

Gusset Material:  Chose 0.060" fiberglass for structural plates (motor mounts and motor-mast to platform plates.  Couldn't find a carbon sheet that cut nice.  Chose 0.040" fiberglass for edging members to protect the sandwich panel from edge loads (e.g., where the zip ties go through the panel.

Joining System:  Nylon bolts, washers, and nuts; UV rated zip ties; high-shear & peel strength epoxy adhesive.

Legs:  Pool noodles or pipe insulation.

Batteries:  Some weight can be saved on batteries.  The weight of a LiPo is dependent on the core material, plastic/tape wrap, wire, and connectors that change as the battery size increases.  If you can choose the lower C rated batteries you will save some weight.  Within the same manufacturer, there is an optimal battery size (found by taking the actual weight / capacity.  For Turnigy, it was the 3S 3000 mAH battery (this might have changed with their new batteries).  If I fly with four 25C batteries, I was able to use the lowest C rating.  This area has room for a lot of weight savings but haven't had time to investigate more that that.    

So with those limitations imposed on myself to meet the requirements, it's now time to design the ship.

Installment 3:  Design

The APM is extraordinarily flexible.  It can handle motor layouts that deviate from regular polygons (for example all of the designs below).  So the question for me was, which irregular polygon shape leads to the lightest weight and stiff frame?  To do this I used an Excel based CAD system that evaluated the following Variant shapes--the primary criteria being the mass of the structure required to support that shape.

First the regular octagon.  Stable platform but no field of view.  It's also heavy.  The CAD system calculates the weight of the black lines, which represent the motor masts.

The first morph was the U with a wide open front.  Now we can meet the FOV requirement.  But as this ship is morphed to a wide enough FOV, it weighs too much even when the masts are optimized into a V-crosshatch layout.

The second morph was the slanted H (commonly called a V).  The H can have parallel masts or masts that taper to the back.  Now in the realm of weight, but can we do better?

The third morph was the C that really compresses the layout.  I really liked this design.  It's light and extremely strong.  But I didn't want to go that extreme without more practical knowledge of how extreme aspect ratios impact flight.  Also note that where thrust is generated by the propellers, that like a regular octagon layout, the props only cross the motor mast once, not twice (and sometimes 3 times) as in the H case above.

The fourth morph was the 8--8 and variants of it (other free-form shapes not shown).  The 8--8 is really getting light, but I was not comfortable yet with a mast bonded to another mast only in one spot.  I planned to crash a lot and needed something that would survive.  

So settled on the X2.

Pros were:

o lightest predicted weight but I'm sure some other design is lighter

o crash survivable with the nylon zip ties designed to break before the motor masts

o easy to initially build with only two crossing members

o quick to assemble in the field (only two masts)

Cons were:

o half of the props go over the frame twice but tested this penalty and it appears to be minor with round tubes

o width (not exactly compact) but still straps onto a backpack

o need to mod the variant code in firmware (true with all irregular polygon motor layouts) but easy to do and the Excel based CAD program generates the required C++ code so all you have to do is paste it into the right place and upload to the copter.

So with the X2 layout, all I had to do was enter the requirements and let the CAD system optimize.  The requirements I entered were:

o 15" props with a 0.1" minimum tip-to-tip gap

o 120 degree field of view with the camera sensor located not more than 7" in front of CG

o a 7.2" open air spacing in the back props for the electronics platform and prop free space

o a prop rotation of -1 (versus +1) for prop spin opposite of conventional

The video shows the CAD system morphing design until the optimal weight is found that meets the requirements.

Tomorrow it gets fun ... I start the build.

Installment IV:  Design Validation and Fabrication

Do these two steps in parallel.  Validating design is making prototypes of complex stuff like the electronics platform and making sure everything fits.  But some of the parts are simple, like the motor mast.  So just make them.

Step 1:  Tape a full-scale print out of the design to the floor.  There are CAD systems for free that will do this and others that don't cost much that also will create cutting profiles for 3-axis CNCs for the mounting plates and gussets.  Make sure that the floor location chosen is flat.

Step 2: Tape the corners of the mounting plates to the design.  Avoiding putting tape where the adhesive goes.  Put tape where you don't want the adhesive to go. 

Step 3: Tape a tool to the floor to hold the motor mast in place while the adhesive is curing (the little bars at the end of the motors masts taped to the floor with duct tape).

Step 4:  Mix the adhesive.  Stir it well as this determines how well it cures.  Shown is Hardman Orange that has high shear strength and peal strength and OK temperature resistance.  Another choice would be 3M Scotch-Weld EC 2216.  The Orange comes in a double bubble pack that you fold in half, cut, and the two parts come out on top of each other nice and neat and ready to stir on supplied wax paper pads (they also supply the stir sticks).

Step 5:  Apply a 3mm or so tall bead down the center of each of the plates (motor mount X plates and motor mast to electronics platform plates) that go on the same side of the motor mast.

Step 6:  Carefully locate the round motor mast tube onto the beads.  Rotate the motor mast back and forth around 12 o'clock (from 11 to 1 o'clock) so that the adhesive can bond to a larger area on the motor mast.  Check that there is an even application of adhesive to the tube and plates and that the outer zip tie holes are free of adhesive.

Step 7:  Tape the tube to the holding tools.  Place weights onto the tube.  Let cure over night.

Step 8:  Remove that motor mast and do the other motor mast.  Let it cure over night.

Step 9:  Flip the motor mast over that has plates on both sides (one will have plates only on one side; see photo below).  Repeat the bonding process for those plates except this time make sure the weights on top of the motor mast hold the top plates parallel to the bottom plates.  An easy way to do this is repeat the clearance to the floor (the part of the motor mast that was cut off plus two fake plates) and balance the weight using this spacer.  Let cure over night.

Step 10:  Cut out the electronics platform from something temporary like thick cardboard.

Step 11:  Get some inexpensive 8" zip ties.

Step 12:  Assemble the ship onto the electronics platform using the zip ties.

Step 13:  Assemble the wire jungle onto the prototype platform.

Step 14:  Validate the electronics platform design.  It usually takes me two prototypes to get everything right.  Make sure that:

o all of the zip tie holes for the wires are located correctly.

o all of the zip tie holes for the motor masts are located correctly.

o all of the wire through-holes are where they should be.

o that there are holes for making PID tuning easier.  the center of the ship isn't where the bars cross, it's at the center of the APM in the above photo.  note the four holes to the left and right of center.  those are for tying the copter down when tuning the pitch PID.  note the two holes fore and aft and equally spaced from center (one is under the read wire cap).  those are for tying the copter down when tuning the roll PID.  There is an additional hole located at CG under the APM.  That is for tuning the yaw PID and for final tuning of pitch and roll after the ship is behaving well enough to trust it indoors on a single bungee chord.

o holes for the APM if you direct bolt it; holes for the GPS; holes for the zip ties for the receivers and other electrical components.

o holes for zip ties to secure the wires running from the ESCs to the motors.

o holes for zip ties used to tie down battery wires

Step 15:  Cut out the electronics platform.  In this case, the electronics platform is made from single play carbon 1/4" 2 pound Nomex core sandwich panel.  On an aircraft, you primarily walk on double carbon ply .4" floor panels that have a higher density core.

Step 16:  Bond the reinforcement plates to the electronics platform.  A single ply of carbon won't handle side loads from small objects like zip ties pulling the motor masts against the platform (zip ties segments are temporarily used to align the plates over the holes during cure) or the Velcro strapped around the batteries (the small rectangular plates).  0.040" fiberglass plate is adequate.  Do one side.  Let cure for 3 hours.  Flip it over and do the other side and let cure over night.

Step 17:  The plates take little adhesive.  Use the rest on the ends of the motor masts.  Put small beads of adhesive on the inside and outside of the carbon ends to help keep the ends together during a crash.  This will help keep single strands or tows will pealing away from the rest of the laminates.  It would be best to bond a round cap made from 0.060 fiberglass to each end.  During final assembly, we will also bond a small piece of foam to further protect the ends.  Let this cure over night.

To summarize, this can take a few nights of cure time.  But while its curing, you can study up on the hard stuff, like the electronics.

Tomorrow we do final assembly and integration.

Installment V:  Final Assembly

o Zip Tie the platform to the motor masts.

o Mount all the electronics to the platform.

   - APM in this case is directly bolted to the upper motor mast at one end and the platform at the other end.

   - Radio receiver is zip tied with the control wires separated from the DC voltage wires. Antenna is pointed forward.

   - All wires are zipped tied to the platform with adequate strain release loops.

   - Control wires that cross DC wires are crossed at right angles.

   - Ratio satellite is zip tied to the platform.  Left antenna points left.  Right antenna points down.

   - GPS is bolted to the platform only close enough to the motor mast to provide protection from an 'inverted landing'.

   - Velcro is added under the ESC locations, power module location, and battery locations.

   - ESC are attached with velcro.  The large DC current wires are widely separated from the ESC control wires.

   - The velcro battery straps are inserted.

o Mount the motors (no props at this point)

o Test the electronics (see documentation for APM build at main wiki)

   - APM lights

   - Radio lights

   - ESCs programmed and throttle set.

o Prepare the prop mount area.  On bolt-mount carbon props, the mount area might not be smooth.  Lightly sand until flat.

o Balance the props Axial - This is done as you've seen many do before using readily available prop balancers.  On carbon props, add tape as needed near the tip.  Do not sand the blade as this can wear through the Epoxy and expose fibers.

o Balance the props Planar - Mount the prop onto the motors.  Lightly tighten the prop.  Spin the prop by hand and measure the distance from the ground at two opposite points.  Make sure that both blades tips trace the same path.  This is the largest source of prop vibration.  On carbon direct mount props, you can usually just tighten the prop on the high side more to bring the blade down.  Sometimes with other types of props, you can simply rotate the blade to a new position under the nut and see if that helps.  If not, remove the prop and appropriately sand the mount area.

That was a good day.  Tomorrow we integrate (make the firmware fly this puppy).

Installment VI:  Integration

Today we load the custom firmware for this unusual motor layout.The reality is that loading custom software is as easy, once the basics are complete, as loading standard firmware.

1)  Create a personal work space for the APM code on your computer (for details, see attached).

2)  Download the software to that workspace.  Currently the code resides at   (for details, see attached).

3)  Find the source code that needs to be modified.  Look under the Ardupilot-ArduCopter-3.0>libraries>AP_Motors directory for the appropriate file.  In this case, for an octa, the file is AP_MotorsOcta.cpp. In that file you will find the paragraph of code that discusses a V(ariant) frame:

    }else if( _frame_orientation == AP_MOTORS_V_FRAME ) {

        // V frame set-up

        add_motor_raw(AP_MOTORS_MOT_1,  1.0,  0.34, AP_MOTORS_MATRIX_YAW_FACTOR_CW,  7);

        add_motor_raw(AP_MOTORS_MOT_2, -1.0, -0.32, AP_MOTORS_MATRIX_YAW_FACTOR_CW,  3);

        add_motor_raw(AP_MOTORS_MOT_3,  1.0, -0.32, AP_MOTORS_MATRIX_YAW_FACTOR_CCW, 6);

        add_motor_raw(AP_MOTORS_MOT_4, -0.5,  -1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CCW, 4);

        add_motor_raw(AP_MOTORS_MOT_5,  1.0,   1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CCW, 8);

        add_motor_raw(AP_MOTORS_MOT_6, -1.0,  0.34, AP_MOTORS_MATRIX_YAW_FACTOR_CCW, 2);

        add_motor_raw(AP_MOTORS_MOT_7, -1.0,   1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CW,  1);

        add_motor_raw(AP_MOTORS_MOT_8,  0.5,  -1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CW,  5);

4) Get the code for your copter.  In the Excel CAD program the code was generated and put into cell P2, the green cell.  Copy that cell (ctrl-C).

5) Paste (ctrl-V) the generated code for your copter into the V-Frame paragraph replacing the old add_motor_raw_lines of code.  Delete or comment out the old add_motor_raw lines. In my case, the code now looks like this:

}else if( _frame_orientation == AP_MOTORS_V_FRAME ) {

// V frame set-up *****original numbers*****
// add_motor_raw(AP_MOTORS_MOT_1, 1.0, 0.34, AP_MOTORS_MATRIX_YAW_FACTOR_CW,7);
// add_motor_raw(AP_MOTORS_MOT_2,-1.0,-0.32, AP_MOTORS_MATRIX_YAW_FACTOR_CW,3);
// add_motor_raw(AP_MOTORS_MOT_3, 1.0,-0.32, AP_MOTORS_MATRIX_YAW_FACTOR_CCW,6);
// add_motor_raw(AP_MOTORS_MOT_4,-0.5,-1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CCW,4);
// add_motor_raw(AP_MOTORS_MOT_5, 1.0, 1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CCW,8);
// add_motor_raw(AP_MOTORS_MOT_6,-1.0, 0.34, AP_MOTORS_MATRIX_YAW_FACTOR_CCW,2);
// add_motor_raw(AP_MOTORS_MOT_7,-1.0, 1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CW,1);
// add_motor_raw(AP_MOTORS_MOT_8, 0.5,-1.0, AP_MOTORS_MATRIX_YAW_FACTOR_CW,5);

// V frame set-up *****Frantz V 7/15/2013*****
// add_motor_raw(AP_MOTORS_MOT_5, 0.945, 1.000, 0.801, 5);
// add_motor_raw(AP_MOTORS_MOT_7, -0.945, 1.000, -0.801, 7);
// add_motor_raw(AP_MOTORS_MOT_1, 0.757, 0.333, -0.801, 1);
// add_motor_raw(AP_MOTORS_MOT_6, -0.757, 0.333, 0.801, 6);
// add_motor_raw(AP_MOTORS_MOT_3, 0.570,-0.333, -0.629, 3);
// add_motor_raw(AP_MOTORS_MOT_2, -0.570,-0.333, 0.629, 2);
// add_motor_raw(AP_MOTORS_MOT_8, 0.383,-1.000, 0.629, 8);
// add_motor_raw(AP_MOTORS_MOT_4, -0.383,-1.000, -0.629, 4);

// X2 frame set-up *****Frantz V 9/22/2013*****
add_motor_raw(AP_MOTORS_MOT_5, 1.000, 1.000, 1.000, 1);
add_motor_raw(AP_MOTORS_MOT_7, -1.000, 1.000, -1.000, 2);
add_motor_raw(AP_MOTORS_MOT_1, 0.546, 0.504, -1.000, 3);
add_motor_raw(AP_MOTORS_MOT_6, -0.546, 0.504, 1.000, 4);
add_motor_raw(AP_MOTORS_MOT_3, 0.379,-0.504, 1.000, 5);
add_motor_raw(AP_MOTORS_MOT_2, -0.379,-0.504, -1.000, 6);
add_motor_raw(AP_MOTORS_MOT_8, 0.834,-1.000, -1.000, 7);
add_motor_raw(AP_MOTORS_MOT_4, -0.834,-1.000, 1.000, 8);

// Test Bench *****Frantz 8/6/2013*****
// add_motor_raw(AP_MOTORS_MOT_1, 0, 0, 0, 1);
// add_motor_raw(AP_MOTORS_MOT_2, 0, 0, 0, 2);
// add_motor_raw(AP_MOTORS_MOT_3, 0, 0, 0, 3);
// add_motor_raw(AP_MOTORS_MOT_4, 0, 0, 0, 4);
// add_motor_raw(AP_MOTORS_MOT_5, 0, 0, 0, 5);
// add_motor_raw(AP_MOTORS_MOT_6, 0, 0, 0, 6);
// add_motor_raw(AP_MOTORS_MOT_7, 0, 0, 0, 7);
// add_motor_raw(AP_MOTORS_MOT_8, 0, 0, 0, 8);
}else {

Note that the // makes what follows it a comment that is ignored during the compile.

6) Save the file.

7) Download the firmware compiler (for details, see attached).

8) Verify the custom firmware code (for details, see attached).

9) Configure the compiler to your work-space (for details, see attached).

10) Compile and upload the custom firmware.  One mouse click does it all! For details, see attached.

11) Set Mission Planner parameters.  Load Mission Planner as normal and connect.  Under Initial Setup > Mandatory Hardware> Frame Type, select V.

You are now ready to test the copter.  Arm and make sure that all the props are rotating in the right direction.  Don't go nuts, there is more to do.

It's been a good day, so tomorrow we'll test and tune.


Installment VII:  Test & Tune

The tune part will take weeks as you try different flight modes.  But if the ship will hover without tuning, take a baseline measurement to see how the copter is doing. Use the free Hover Analysis worksheet to set your baseline so you will know how much the ship is improving during the tuning process.  I'll post updates in the Vibration blog.  See the vibration wiki to see how to enable IMU/RAW, GPS, ATT, and CTUN, which are the records that Hover Analysis needs to analyze vibration and stability.

For the X2 I saw this.

The left chart shows vibration at the APM.  The short horizontal bars is where the system found the most stable 12 to 20 seconds of hover to analyze.  That is why we have the ship hover for two minutes.  Inexperienced pilots like myself need two minutes to get 12 seconds of stable hover!  To fly, you need each random acceleration (vibration) to be less than .5 gravities (G).  The numbers in blue, purple, and gray above the chart show the metrics for each axis.  And above that is the average.  The X2 has an average of 0.06.  That is considered excellent.  This is especially good since the craft is lacking its normal payload of around +2kg in batteries and camera.  Fully loaded, those numbers would drop significantly.  This metric is a 99.5 percentile number of the absolute values of the raw data ... a vibration spike that the APM sees every 200 observations (necessary because the APM only samples accelerations).

Keep in mind that for the X2, the APM is bolted (no dampeners or isolators) directly into the same mast used to hold four of the rotors.  The other four rotors are zip tied to the same platform.  The point being that if you:

o balance your props in x/y (on a standard balance)

o balance your props in z (the tip trace; the most intense source of vibration in z)

o use a stiff motor mast (decreases acceleration)

o reflect vibrations off of higher masses (where the motor masts meet the zip tie zone holding the central node).

o dampen by tying the mass directly to to APM ...

the ship won't need isolators for the extremely light APM.  However, there are many ships that may need isolators.  And, the camera gimbal can almost always benefit from isolation/dampening.

Another number to watch is the signal to noise ratio on acceleration.  Try to keep this above 2.  The X2 is at 4.  If I hadn't met the design goal of > 3 then more or larger zip ties could be added to the motor masts to tame vibration.

The right graph shows stability.  As you dampen, stability won't change much (as long as the PIDs are tuned for the weight).  But if you isolate, stability will get worse.  Consider that a law of physics.  In stable mode at hover, yaw is not controlled, so the statistics only looks at roll and pitch.  The metric is in degrees from level and is the standard deviation from level.  The X2 baseline (without tuning) had a stability of .22 and .17 respectively.  That is good, but there is room for improvement, so the X2 does need PID stability tuning.

So let's tune.

Some like to tune outdoors in flight using the variable knob on the radio.  Some like to tune indoors.  It's a personal preference.  I tune indoors and then validate outdoors.  I have a slightly unique approach that uses Mission Planner to track pitch and roll after a change in a parameter.  I count the number of observations > .3 degrees per 50 and modify the rules as I hone in on the best reading.

Now rerun the Hover Analysis to see if you made improvement.

Much better!  Even the Signal to noise increased to 4.3.  Roll and pitch are now both in the excellent range.

Also in the Hover Analysis worksheet, in the orange area, is the THR_MID number that needs to be entered into the Standard Params in Mission Planner.  Enter that value into the Throttle Mid Pos.  This will help keep the ship hovering at about 50% throttle.  I'll recheck that number for various payloads.

Congrats!  You now have a flying machine that is better than most commercial ships.

Another good day.  Tomorrow we validate.  That is where we match up copter performance to requirements.  Did we make it?  Tomorrow we find out.

Propellers at Different Levels?  Does it matter.  Was asked this question.  So am putting the question out there for comment.

History:  There are many commercial examples of helicopters with multi-rotors at different levels." alt="Photo ID: 1336036 Views: 14628 UK - Air Force Boeing Chinook HC3 (352) (ZH898) shot at Off-Airport - Salisbury Plain Training Area UK - England March 7, 2008 By Karl Drage - Global Aviation Resource"/>

Reasons:  Better prop tip separation (less noise/cleaner air).

Impact on lift:  None to my knowledge

Impact on control:  Let's go to extremes.  Take two sets of rotors 10 meters apart:  two on the same level and two stacked on top of each other.  For the two at the same level you roll capability.  For the two stacked on top of each other you have no roll capability.  For the same level case, as air varies around each prop, roll can be induced creating instability.  For the stacked, roll isn't induced so the ship remains stable.  So from an extreme point of view, do you want stability or roll/pitch response?

Mathematical difference:  This is all controlled by the aspect ratio of the propeller plane.  With a 'same level' design, the aspect ration of height to width is zero.  The aspect ratio of the X2 is .023 or a 2.3% difference in roll and a 4% difference in pitch.  This would have a small impact either way on increased stability or decrease in roll response. However, if this was an acro machine build for competition, then I'd look hard at a different frame approach that puts all the motors at the same level.  The X2 is built for photography.

Intallment VIII:  Validation (videos) of the X2 Black Momba (motor loss test)

Does it do what was intended?  Let's go through each requirement.

FOV:  An 80 degree clean (no props) field of view for a camera sensor located at the ship geometric center (full size camera).   A 120 FOV with the camera sensor located less than 7" forward of center (go pro) for pan 90 degree downward to forward to 90 degrees upward.

Yes.  Done via design.  The front is wide open.

Total frame weight (bolts, masts, spars, platform, gussets, ties, etc.) < 400 grams.

Close.  407 grams.

x/y/z accels < .1G

x/y/z accels all below .08G with average at .06G.

Not fail at 9G (crash) x 3.2 kg (approx max ship load)

TBD - not brave enough to test yet

Land safely with any 1-motor failure

Yes.  Lands and takes off with the outside or inside rotor failure (see video of inside motor failure).  The outside test wasn't filmed.  It happened by accident.  I'd taken off, not paying attention, when my brother said, "Hey, think you better land.  You lost a motor."  It was flying so stable that I didn't even notice (not good pilot technique; I have a lot to learn).

Rotors with > 2 x 2.3 kg net-thrust (thrust after lifting themselves) or 575 net grams thrust per rotor.

Yes.  This was accomplished during the selection process of testing motor efficiency.  Net thrust for each rotor was 792,779,766,778,802,779,772,785 grams for an average of 782 (in red italic below).

APM located at center of x/y CG and positioned at the propeller z-plane.

Yes.  Per design.  There is a central hook centered under the APM that allows the pilot to pick up the craft from a single dead center point and balance the ship by moving around the payload (batteries and camera).

Able to backpack.

Yes.  Comes apart by:

- cutting 6 zip ties

- fold and velcro the two motor masts together

- attach to split-board (snow board) backpack using same method

- also bring zip-tie gun, zip ties, and zip-tie cutter (along with flying equipment and checklist)

- hike to location

- remove from backpack

- remove holding velcro and locate arm onto electronics platform lining up the plate holes

- apply six zip ties using zip-tie gun

- check out all electrical and mechanical connections

Able to assemble in field in < 12 minutes.

Yes.  At a relaxed methodical pace it takes:

o 3 minutes to disassemble and velcro the masts together ready for transport.

o 10 minutes to remove from the pack, relocate/tie the masts in place, check all wires and mechanical connections.

Flight time with GoPro of > 30 minutes.

TBD.  But did flight test using 12000 mAh batteries.  The ship will carry a Go-Pro Black with 15000mAh batteries and still weigh less than the 50% of lift capability limit.

Looks promising.  Flight duration on only 95% of 12000 mAh was over 30 minutes (no gopro but the gopro black only weighs 10 grams per motor).  With 15000 mAh the flight would be approximately 36 minutes (power consumption would increase from 22.1 to about 25 amps with the go-pro and extra battery capacity).  So appears to have adequate margin.

See the following video of the duration test.  The log file is attached (was not tuned for for the payload).

Ability to support gimbal for high quality stills and videos.

TBD, but adequate mounting space on front.  Can cut new electronics platform if needed.

140 Hz < hover frequency < 180 Hz.

Partial success & failure.  Depending on payload, between 175Hz (lightest) and 212Hz (normal payload) Hz.  But will be able to test 200 Hz flight by adjusting the payload and seeing the impact on stable hover.

That's it.  Didn't take too long to build.  Now I'll finish a few things up like tuning for different payloads; research batteries and break the 30 minute barrier; and start working on the camera system.  Should be fun.


Wow! Excellent post. You should move it from unassigned category to arducopter group category because everybody like me missed it. 

How would you plan to fix your gimbal on this frame?

Why not installing the motors upside down and reverse the props, so that the props do not blow on the beams? This would force a landing gear design allowing this.

What would be nice , but probably a dream, would be to integrate the battery within the hollowed carbon tubes...

The way you stack ESCs is not ideal for heat dispersion because you block convection and radiance. But it depends if they heat up much or not after 30min.

APM case is not protected from the sun: heat and light. The integrated barometer is sensitive to light and heat. The translucid case and foam on top of the baro is not enough. Either a black dome on top, either paint the case black. I experienced, like others issue with APM in full sun.

going to sleep now, it is 2am...will comment further with fresh eyes...

Great puts.  Didn't know about the sun light.  that might explain why my V had alt-hold issues.  would sit there for a while and then just slowly fall out of the sky.

that battery idea is such a great idea. wish i knew how to make them.  i could save another 150 grams probably.

with the batteries on the bottom, the camera is going out front at frame level.  when i do this, one or more batteries will move to the back.  this will give the camera the ability to point straight down, out, or up.  i live by a volcano (mostly inactive) and have this image of the camera pointed straight up as it takes off straight up through the trees and then as it peeks out even with the trees (around 40 meters), it rotates flat to the mountain, flies over the waterfall and then pans down to the abyss.  that is the image that has driven me on this project.

actually had the motors reversed (lawn mower style) on the V for that reason.  but then did some tests and found that to my great surprise that it had little impact on vibration and only a small impact on lift.  really strange.  doesn't make sense.  the round tubes help, but ???

so would putting the 18-amp esc's on end with the hot side up help?  i only run at less than 3 amps which is why they haven't been a problem, but i need to be setting a good example, which i'm not.

i'll find a cheese container or something and put it over the APM.  should i paint it silver, white, or black?


so is there a way to move it to the unassigned category?

time for dinner.  2am huh?

Reply to Discussion



Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2017   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service